↖ Back

Generative AI and Vertical SaaS – Where Things Stand  

In March 2023, I wrote a post about the potential impact of generative AI on vertical SaaS. Since then, the generative AI world has remained red hot at every level, from chips and models to middleware and applications. With all that momentum, I thought it would be a good time to check back in on how things are playing out in terms of vertical SaaS. This post has two sections. In the first half, I highlight some key trends at the application and company level. In the second half, I try to frame some of the strategic dynamics at play in the market.

Key Trends in Generative AI for Vertical SaaS 

1. Generative AI is changing the way we work, slowly 

In last year’s post, I argued that internal company operations are the low hanging fruit for integrating generative AI. Based on what we see in our portfolio, pipeline, and internally, this prediction has borne out, though perhaps not as quickly as expected. For example, software engineers are a key vector for the spread of AI copilots, but not all of our portfolio company engineering teams have embraced them. Adoption of AI first marketing tools has also been mixed. In the case of ChatGPT, it’s clearly a versatile tool for search, inspiration, and data transformation. But personally, it has only taken over a small portion of my workflows. Some internal automation areas are booming, especially customer service. But for generative AI to “double the growth rate of ARR per employee“, the technology needs to become more integrated into the products that we use to work. Which of course is the topic of this post!

2. Generative AI is shifting roadmaps for in-market companies 

Injecting generative AI into existing products is tricky. Which roadmap item gets bumped? How do you skill up the team? Will the accuracy and explainability meet your customer’s expectations? In many cases, however, the massive customer value that can be unlocked means companies have had no choice but to reprioritize.

At the top of the market, vertical SaaS incumbents have shown agility in pivoting toward generative AI. Just like large horizontal players such as Adobe, Notion, and Figma, vertical SaaS leaders like Five9 (call center software), Procore (construction tech) Clio (legal tech), and Epic (medical records) have leveraged their large engineering teams, proprietary data pools, and A+ model access to bring generative AI into their products quickly. It’s too soon to know much about usage and impact, but at least for round one, the big players are setting the pace.  

Earlier stage companies are also proving nimble. One typical pattern is companies with rich data sets launching a co-pilot experience to enable new forms of access and interaction. For example:

  • MinuteBox, a corporate records company, launched Second Chair, an AI powered tool to interact with those records.
  • Alexi, another Canadian legal tech company, launched Arguments, an LLM powered brainstorming and decision making tool that sits atop its core case referencing engine.
  • Pathway Medical, a diagnostic resource for doctors, launched a co pilot to enable chat interactions.

Another pattern is to layer an AI-first feature on top of an existing system of record:

  • Our portfolio company Jane Software has integrated with Clinic Sites, an AI-first clinic website generator that is tied into Jane’s core scheduling and clinic management tools.
  • Another portfolio company, Distiller SR, has launched an LLM powered data extraction tool within its literature review product.
  • Mikata Health, a leading Canadian clinic management software, launched a scribe product on top of its clinic management product.
3. Generative AI first companies are coming fast 

While in-market companies transform their roadmaps, a wave of generative AI-first vertical SaaS companies are also gaining traction. The world of health tech, a focal point for our investment thesis, provides an interesting cross section of approaches:   

  • Ship a best-in-class featureNabla and Abridge are well funded startups building scribe products that integrate into the larger EMR and clinic management landscape. In Canada, Tali and Mutuo are pursuing a similar path.  
  • Ship all the featuresAmbience and Fabric both position themselves as “operating systems” integrating generative AI tools across a range of patient engagement tasks, from intake through follow up.  
  • Verticalize a feature that is working elsewhereHyro AI builds AI powered customer service bots focused on healthcare.
  • Reimagine a core use case Untether Labs is building AI first staff scheduling.
  • Start “selling the work” – In health care, selling the work means wading into diagnosis and treatment, an exciting but complicated proposition. Early examples include AI Berry, for mental health diagnosis, Glass Health for diagnosis and clinical planning and Cass, a mental health chatbot.   

Generative AI first companies are coming fast to other verticals as well: 

  • Legal tech is one of the hottest areas. Spellbook is a leading Canadian player, their contracts product grew 10X in 2023. Harvey AI is the US fundraising leader in the space.
  • Permit Flow is well financed AI first construction permitting software.
  • Companies like Hypar and Maket are bringing generative AI tools to builders and interior designers.

Competitive Dynamics in Gen AI for Vertical SaaS 

Zooming out from companies and products, it’s always worth asking about competitive forces. What strategies will be optimal to win in the gen AI era? And which type of companies are best positioned today? Here are a few thoughts from that perspective. 

1. Incumbents have real advantages 

Just like in infrastructure and horizontal apps, vertical SaaS incumbents have meaningful advantages integrating generative AI, as their growth investors will be happy to tell you. Incumbent products are established control points, sticky products with inherently low churn. Their customers are technology laggards who prefer to buy vs. build. Market leaders have unique data sets, deep customer insight and best in class customer support. While my job is to invest in disruption, I do find these arguments compelling to a point. The best run enterprise vertical SaaS companies should be able to use generative AI to increase customer lock in while growing ARPU as software takes a greater share of human tasks.   

2. Startups are coming hard

In the last two months, name brand VCs announced 300M of funding for various takes on generative AI + health care documentation. No doubt there are more than that, not to mention all of the seed funded, bootstrapped, regional players and so on. Similar races are on in other hot verticals, and the pace seems unlikely to slow. Startup activity and venture investment don’t guarantee industry transformation. Given startup’s advantages in speed and focus plus the platform level disruption taking place with gen AI, however, it seems like a safe bet that at least some of today’s AI first solutions will use their technology wedge to build sticky systems of record that put them in the top echelons of vertical SaaS winners.  

3. Innovating at the data layer will be crucial

For generative AI first disrupters, innovating at the data layer will be crucial to creating competitive moats. Without unique data sets and custom trained models, new entrants can be easily copied by incumbents or other competitors. On the other hand, startups that amalgamate the talent, data and compute to create their own models are doing something that is difficult, expensive and much harder to copy. It’s also likely out of scope for the large horizontal model providers. It’s no coincidence that Abridge, the best funded of the clinical documentation startups noted above, “plans to use the fresh funding for hiring as well as acquiring and training purpose-built medical AI models.” For more on the topic of data moats in generative AI for vertical SaaS, check out this article    

4. AI-Second can still be a great strategy 

After 15 years, market penetration of cloud based VSaaS is still nascent. This means that many vertical use cases have still not been digitized in a modern way. We see this all the time in our pipeline. Most companies we meet are not running a generative AI first playbook, they are running a traditional vertical SaaS or marketplace playbook, finding customer problems and solving them with modern software. And that makes perfect sense. AI second companies should approach generative AI like any good in market company, winning with their core value proposition while looking for ways to inject the tech into their operations or their products where it makes sense.   


Over the past year, generative AI has continued its breakneck progress toward becoming the next layer of the software stack. Like cloud or mobile, not only will all software systems have access to a whole new set of capabilities thanks to generative AI, but those companies will need to leverage these capabilities to deliver full value to their users.

While generative AI is transforming the stack, the broader game for startups or incumbents remains the same – attract the best people, build a great culture, listen to your customers, and ship value. AI first or second, great software is hard to build and sell, but great teams figure out how to build and sell it. As investors, our focus remains squarely on finding these teams to work with.   

Reading List

Back to top ↖